| article: Planetary Imaging - Getting more Detail with Multiple Unsharp Mask |
The History
The term "Unsharp Mask" was invented during processing and copying of film images.
An unsharp image can be regarded to be a sharp one with an overlayed unsharp one.
So why not to subtract the unsharp part of the image?
It was very tedious in former times to prepare an accurate unsharp mask and subtracting it the analog way.
Digital Images and Standard Unsharp Mask Filter
A lot of image processing tools do have a filter called "unsharp mask".
All the ones I have tested seem to be nothing but sharpening filters such as high pass filters.
The simulation of the real unsharp mask is going well beyond these filters.
It is more flexible, more powerful and can be used with different blur in the unsharp mask, hence the name "multiple unsharp mask".
The idea is to prepare several blurred masks of the original image with different amount of gaussian blur to reflect the atmosphere's blurring by seeing.
If you want to know more about seeing check the article on atmospheric effects in astronomy.
  
Mars at 75 Degrees from Zenith - before unsharp masking, after the first and after the second unsharp masking
  
Jupiter at 70 Degrees from Zenith - before unsharp masking, after the first and after the second unsharp masking
The seeing was so bad that these images cannot be restored to perfection.
But the amount of more detail visible is still amazing.
Simulating the traditional Unsharp Mask Process
I will describe the method for AstroArt, but a lot of other image processing programs will do the job.
What the software must be able to do is basically
- Duplicate an image
- Apply a variable blurring filter (softening filter)
- Subtract two images from each other with a variable multiplication factor (image weight)
In AstroArt you would have to do the following to prepare the first mask:
- Open your image
- Call from the menu Image / Duplicate
- Call from the menu Image / Duplicate again to be able to have "2-stage unsharp mask"
- The first duplicated image will be blurred by Filters / Gauss. The value Sigma is very much depending on the size of you image. If you supersampled your image during combination of the raw frames start with 4.0 otherwise with 2.0
The resulting image should have significantly more blur than your original image:
The gaussian blurred images will act as the unsharp mask
In AstroArt you would have to do the following to perform the unsharp mask subtraction:
- Select your original image
- Call from the menu Arithmetic / Merge
- Select the name of your first unsharp mask (maybe NoName00.fit)
- Set the values to the following:
- Just press PREVIEW
- Watch the original image. The factor 1.1 is the important one. If the original image is way too dark or way to bright change this value in small steps
When you are satisified click OK. Fine tune the bright and dark areas with the View / Histogram by dragging the two horizontal sliders.
The resulting image should have significantly more detail than your original image.
Repeat the both steps but now prepare your mask with 2 times the value of Sigma for the blur as chosen before.
Again the image will have more details but on a different scale (or frequency).
Try to understand whith which value of Sigma you can start with your particular seeing conditions and equipment.
This might mean a lot of trial and error but it will make the final difference.
Depending on the signal to noise ratio in your image you can repeat this process several times.
The less noise your image had the more often this filtering can be applied.
That is why you should have had as many good raw frames averaged as possible to prepare the original image.
If your image now should have strong red and blue fringes at the edge of the planetary disk please check this article: Atmospheric Dispersion Correction
|
|
|